Abstract
BACKGROUND: "Secondary" atrial fibrillation (AF) denotes AF that is precipitated by short-term triggers and that may be reversible. Using administrative data to study secondary AF is of interest, but the ability of these data to verify secondary AF has not been studied. METHODS: We conducted a cross-sectional analysis of 1000 randomly selected hospitalizations of patients discharged alive between January 1, 2016 and March 31, 2020, with AF coded as the most responsible diagnosis (type 1), post-admit comorbidity (type 2), or secondary diagnosis (type 3). We compared diagnosis types to AF category (secondary or not) as determined by a physician blinded to the discharge diagnosis type. We calculated the positive predictive value (PPV) of the designation of secondary AF in comparison to physician determination. RESULTS: A total of 421 hospitalizations had AF documented as a type 2 diagnosis; this had a PPV of 94.8% for physician determination of secondary AF. After excluding hospitalizations with preexisting AF, and those for which AF type could not be determined by the physician, the PPV of a type 2 diagnosis (n = 391) for secondary AF was 99.7%. Type 3 diagnoses of AF (n = 222) mostly captured hospitalizations with preexisting AF (87.8% of type 3 diagnoses). CONCLUSIONS: A type 2 diagnosis can be used to verify secondary AF in people who were first diagnosed with AF while hospitalized for other causes. This verification facilitates cohort studies and clinical trial recruitment of people with this AF subtype, although it should not be used to determine the prevalence or incidence of secondary AF.